September 2005
Volume 5, Issue 8
Vision Sciences Society Annual Meeting Abstract  |   September 2005
Rapid animal detection in natural scenes: Critical features are local
Author Affiliations
  • Felix A. Wichmann
    Max-Planck-Institut für Biologische Kybernetik, 72076 Tübingen, Germany
  • Pedro Rosas
    Max-Planck-Institut für Biologische Kybernetik, 72076 Tübingen, Germany, and Centro de Neurociencias Integradas, Facultad de Medicina, Universidad de Chile, Independencia 1027, Santiago, Chile
  • Karl R. Gegenfurtner
    Allgemeine Psychologie, Justus-Liebig-Universität Gießen, 35394 Giessen, Germany
Journal of Vision September 2005, Vol.5, 376. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Felix A. Wichmann, Pedro Rosas, Karl R. Gegenfurtner; Rapid animal detection in natural scenes: Critical features are local. Journal of Vision 2005;5(8):376.

      Download citation file:

      © ARVO (1962-2015); The Authors (2016-present)

  • Supplements

Thorpe et al (Nature 381, 1996) first showed how rapidly human observers are able to classify natural images as to whether they contain an animal or not. Whilst the basic result has been replicated using different response paradigms (yes-no versus forced-choice), modalities (eye movements versus button presses) as well as while measuring neurophysiological correlates (ERPs), it is still unclear which image features support this rapid categorisation. Recently Torralba and Oliva (Network: Computation in Neural Systems, 14, 2003) suggested that simple global image statistics can be used to predict seemingly complex decisions about the absence and/or presence of objects in natural scences. They show that the information contained in a small number (N=16) of spectral principal components (SPC)—principal component analysis (PCA) applied to the normalised power spectra of the images—is sufficient to achieve approximately 80% correct animal detection in natural scenes.

Our goal was to test whether human observers make use of the power spectrum when rapidly classifying natural scenes. We measured our subjects' ability to detect animals in natural scenes as a function of presentation time (13 to 167 msec); images were immediately followed by a noise mask. In one condition we used the original images, in the other images whose power spectra were equalised (each power spectrum was set to the mean power spectrum over our ensemble of 1476 images). Thresholds for 75% correct animal detection were in the region of 20–30 msec for all observers, independent of the power spectrum of the images: this result makes it very unlikely that human observers make use of the global power spectrum. Taken together with the results of Gegenfurtner, Braun & Wichmann (Journal of Vision [abstract], 2003), showing the robustness of animal detection to global phase noise, we conclude that humans use local features, like edges and contours, in rapid animal detection.

Wichmann, F. A. Rosas, P. Gegenfurtner, K. R. (2005). Rapid animal detection in natural scenes: Critical features are local [Abstract]. Journal of Vision, 5(8):376, 376a,, doi:10.1167/5.8.376. [CrossRef] [PubMed]
 Supported by the Max Planck Gesellschaft (MPG) and the Deutsche Forschungsgemeinschaft (DFG)

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.